Molecular Basis of Coxsackievirus A10 Entry Using the Two-In-One Attachment and Uncoating Receptor KRM1

Molecular basis of Coxsackievirus A10 entry using the two-in-one attachment and uncoating receptor KRM1

Yingzi Cuia,b, Ruchao Penga, Hao Songa,c, Zhou Tonga,d, Xiao Qua, Sheng Liua, Xin Zhaoa, Yan Chaia, Peiyi Wange, George F. Gaoa,b,1, and Jianxun Qia,b,1

aChinese Academy of Sciences (CAS) Key Laboratory of Pathogenic Microbiology and Immunology, Institute of Microbiology, Chinese Academy of Sciences, 100101 Beijing, China; bSavaid Medical School, University of Chinese Academy of Sciences, 100049 Beijing, China; cResearch Network of Immunity and Health, Beijing Institutes of Life Science, Chinese Academy of Sciences, 100101 Beijing, China; dShanxi Academy of Advanced Research and Innovation, 030032 Taiyuan, Shanxi, China; and eDepartment of Biology, Southern University of Science and Technology, 518055 Shenzhen, China

Contributed by George F. Gao, May 22, 2020 (sent for review March 24, 2020; reviewed by Yao Cong and Félix A. Rey) KREMEN1 (KRM1) has been identified as a functional receptor for interacts with Dickkopf1 (DKK1) and lipoprotein receptor-related Coxsackievirus A10 (CV-A10), a causative agent of hand-foot-and- protein 6 (LRP6) (8). The structure of the KRM1-DKK1-LRP6 mouth disease (HFMD), which poses a great threat to infants globally. ternary complex shows that DKK1/LRP6 bind to KRM1 at the However, the underlying mechanisms for the viral entry process N-terminal KR and WSC domains without contacting the are not well understood. Here we determined the atomic struc- C-terminal CUB domain (7). It is unclear which domain of KRM1 tures of different forms of CV-A10 viral particles and its complex is responsible for CV-A10 interaction. with KRM1 in both neutral and acidic conditions. These structures A previous study has demonstrated that CV-A10 binds to the reveal that KRM1 selectively binds to the mature viral particle ectodomain of KRM1 at the cell surface and subsequently induces above the canyon of the viral protein 1 (VP1) subunit and contacts endocytosis (6), suggesting a two-step entry process and a probable across two adjacent asymmetry units. The key residues for receptor pH-dependent uncoating mechanism for CV-A10 infection. Struc- binding are conserved among most KRM1-dependent enteroviruses, suggesting a uniform mechanism for receptor binding. Moreover, the tures of CV-A10 viral particles in different states (i.e., mature binding of KRM1 induces the release of pocket factor, a process virion, metastable A-particle intermediate, and empty capsid) accelerated under acidic conditions. Further biochemical studies con- have revealed conserved features for typical enteroviruses,

firmed that receptor binding at acidic pH enabled CV-A10 virion wherein a lipid molecule termed the “pocket factor” is accom- MICROBIOLOGY uncoating in vitro. Taken together, these findings provide high- modated within a hydrophobic pocket underneath the canyon of resolution snapshots of CV-A10 entry and identify KRM1 as a viral protein 1 (VP1) subunit. This pocket factor has been two-in-one receptor for enterovirus infection. Significance KRM1 | Coxsackievirus A10 | attachment/uncoating receptor | viral entry | enterovirus CV-A10 is one of the major causative agents of hand-foot-mouth disease, an acute febrile disease occurring mainly in children. To nteroviruses are a group of nonenveloped positive-sense date, there are no effective vaccines or therapeutic drugs available ERNA viruses belonging to the Enterovirus genus within the for CV-A10 infection. Moreover, knowledge of the mechanisms of family Picornaviridae. Coxsackievirus A10 (CV-A10) is a member CV-A10 infection and pathogenesis, which are crucial for preven- of the type-A Enterovirus (EV-A) subgroup. CV-A10 infection is tion and treatment of the related diseases, is incomplete. In this frequently reported to cause hand-foot-mouth disease (HFMD), work, we provide comprehensive mechanistic insight into CV-A10 an acute febrile disease characterized by fever and vesicular exanthema, infection using the receptor KRM1, demonstrating that KRM1 acts mostly in the hands, feet, and oral mucosa, posing a tremendous as a two-in-one attachment and uncoating receptor for CV-A10 threat to the health of young children globally (1). In addition entry, which mediates both viral attachment at the cell surface and to HFMD, CV-A10 infection can cause a wide range of clinical uncoating in the endosome. This work expands our understanding manifestations, such as herpangina, aseptic meningitis, and viral on the mechanism of nonenveloped virus infection and reveals meningitis (2). Attention should be paid to emerging CV- new clues for antiviral intervention. A10–associated HFMD and its delayed symptoms to prevent potential outbreaks and improve medical care. Author contributions: G.F.G. and J.Q. designed research; Y. Cui, X.Q., S.L., and Y. Chai performed research; Y. Cui, R.P., H.S., P.W., and J.Q. analyzed data; and Y. Cui, R.P., H.S., Most enteroviruses enter host cells through a two-receptor Z.T., X.Z., G.F.G., and J.Q. wrote the paper. mechanism, with the first receptor enabling virion attachment Reviewers: Y.C., Chinese Academy of Sciences, Institute of Biochemistry and Cell Biology; at the cell surface and a second uncoating receptor triggering and F.A.R., Institut Pasteur. conformational changes of the viral particle and facilitating genome The authors declare no competing interest. release into the cytosol (3, 4). However, some enteroviruses, such as Published under the PNAS license. poliovirus, use only one receptor to accomplish both steps for cell Data deposition: The density maps have been deposited to the Electron Microscopy Data entry (5). Kringle-containing transmembrane protein 1 (KRM1) Bank with accession nos. EMD-30253 (mature CV-A10, pH 7.4), EMD-30254 (mature CV- was recently identified as an essential entry receptor for CV-A10 A10, pH 5.5), EMD-30287 (A-particle CV-A10, pH 7.4), EMD-30290 (A-particle CV-A10, pH 5.5), EMD-30291 (empty CV-A10, pH 7.4), EMD-30292 (empty CV-A10, pH 5.5), EMD-30259 and a major subset of EV-As (6); however, the exact molecular (CV-A10/KRM1, pH 7.4), and EMD-30260 (CV-A10/KRM1, pH 5.5). The coordinates of the mechanism of KRM1 for mediating the entry of these viruses corresponding atomic models have been deposited to the Protein Data Bank with ID remain incompletely understood. codes 7BZN (mature CV-A10, pH 7.4), 7BZO (mature CV-A10, pH 5.5), 7C4T (A-particle KRM1 is a ubiquitous membrane-anchored protein located at CV-A10, pH 7.4), 7C4W (A-particle CV-A10, pH 5.5), 7C4Y (empty CV-A10, pH 7.4), 7C4Z (empty CV-A10, pH 5.5), 7BZT (CV-A10/KRM1, pH 7.4), and 7BZU (CV-A10/KRM1, pH5.5). both cell surface and intracellular membranes. The ectodomain The sequence of the CV-A10 HB09-035 strain has been deposited in the GenBank data- of KRM1 consists of three similarly sized structural domains: an base (accession no. MT263729). N-terminal Kringle (KR) domain, a poorly folded WSC domain 1To whom correspondence may be addressed. Email: [email protected] or [email protected]. in the middle, and a pseudo–Ig-like CUB domain at the C terminus This article contains supporting information online at https://www.pnas.org/lookup/suppl/ (7). In previous studies, KRM1 has been extensively characterized doi:10.1073/pnas.2005341117/-/DCSupplemental. as a regulatory ligand of the Wnt/β-catenin signaling pathway that First published July 20, 2020.

www.pnas.org/cgi/doi/10.1073/pnas.2005341117 PNAS | August 4, 2020 | vol. 117 | no. 31 | 18711–18718 Downloaded by guest on September 26, 2021 suggested as an essential component to maintain the stability of particles and the complex structures of a mature virion bound to the infectious mature viral particles, and the release of pocket the KRM1 receptor at both neutral and acidic pH. All these factor is a hypothetical trigger to initiate the uncoating process (4, structures were determined at atomic or near-atomic resolution, 9). How the uncoating of CV-A10 is triggered and whether KRM1 which allowed us to dissect the detailed molecular events during is involved in this process are unknown, however. virion–receptor interactions (SI Appendix, Figs. S2–S4). In this work, we present biochemical evidence indicating that Similar to other enteroviruses, the CV-A10 preparation is mature CV-A10 virions can bind to KRM1 in both neutral and composed mainly of mature viral particles (>80%) that are sta- acidic environments. Structural studies identify KRM1 as a ble under both neutral and acidic conditions (Fig. 2 A and D). In two-in-one receptor for CV-A10 entry and suggest the essential addition, small fractions of A-particles (∼3%) and empty viral roles of low-pH conditions and receptor binding to initiate CV- capsids (∼10%) were also observed, for which a clear open A10 uncoating. These findings provide substantial insight into channel was formed at the two-fold axis (SI Appendix, Fig. S2). the enterovirus infection process and identify new targets for Within the mature viral particle, a lipid molecule was accom- antiviral intervention. modated in a pocket beneath the canyon of the VP1 subunit, a signature “pocket factor” of many enteroviruses critical for virion Results stability (Fig. 2 A and D) (1, 9, 10). Compared with mature viral KRM1 Binds to Mature CV-A10 Virions in Both Neutral and Acidic particles, the empty capsids and A-particle intermediates display Conditions. We propagated and purified CV-A10 from human conspicuously expanded capsid radii, and the VP4 subunit is rhabdomyosarcoma (RD) cells. Two bands were observed on a dissociated from the capsid layer (Fig. 2). Furthermore, the 15 to 45% (wt/vol) sucrose gradient cushion. The top band was pocket within VP1 is collapsed due to conformational changes composed of empty particles, and the bottom band consisted around the canyon, a result of pocket factor ejection to facilitate mainly of mature virions (SI Appendix,Fig.S1A and B). As shown uncoating (Fig. 2). These observations are consistent with a by surface plasmon resonance (SPR) assays, the ectodomain of probable universal mechanism for enterovirus uncoating as ob- KRM1 (residues 23 to 373) selectively bound to mature CV-A10 served for poliovirus, Echo viruses, and other related enteroviruses, although the triggering mechanisms for uncoating vary virions with high affinity (KD = 42 nM) at pH 7.4 (mimicking neutral physiological conditions) (Fig. 1A), whereas the empty among different viruses (3, 4). The structures of free CV-A10 particles in different states particles showed no discernible binding (Fig. 1B). This observation (mature virions, A-particle intermediates, and empty capsids) at supports the role of KRM1 as an attachment receptor at the cell pH 5.5 are highly similar to those of their counterparts at pH 7.4, surface and suggests that empty viral particles may be noninfectious. indicating that free CV-A10 viral particles themselves are stable Because many enteroviruses enter host cells through the under different pH conditions (Fig. 2). In the structure of KRM1 endosomal pathway and uncoat under acidic conditions, we also bound to CV-A10 at pH 7.4, the lipid was well accommodated, tested the binding between KRM1 and mature CV-A10 particles K = albeit with slightly weaker density compared with that in the free at pH 5.5. Despite displaying a slightly lower affinity ( D 210 nM) virion, suggesting reduced occupancy due to pocket factor re- compared with that under neutral conditions, KRM1 could also lease in some particles (Fig. 3B). Interestingly, the lipid mole- efficiently bind to mature CV-A10 virions in an acidic environment, cules in KRM1-bound viral particles at pH 5.5 became almost suggesting that it may also serve as an uncoating receptor in the C D invisible in the density map, implying a more profound pocket endosome (Fig. 1 and ). factor release under acidic conditions (SI Appendix, Fig. S4). These results support the role of KRM1 in mediating CV-A10 Structures of CV-A10 Viral Particles and Its Complex with KRM1. To attachment at the cell surface and imply that it may also serve as understand how KRM1 mediates CV-A10 entry, we determined a low-efficiency uncoating receptor at neutral pH, which would the structures of three different types of free CV-A10 viral become more efficient in the acidic environment. Thus, KRM1 might be a two-in-one receptor for CV-A10 viral entry, responsible for both virion attachment and uncoating.

Structural Adaptation for KRM1 Interaction. In general, the structure of the KRM1-bound viral particle highly resembles the free mature virion; however, several loops at the binding interface display some moderate local conformational changes on receptor engagement (Fig. 3C). The EF-loop in the VP2 subunit, as well as the C terminus and GH-loop of the VP3 subunit, are pulled upward from the viral surface to enable close contact with the receptor (SI Appendix, Figs. S5 and S6). Of note, a long loop (termed the gating loop) within the canyon of the VP1 subunit (around residues 210 to 230) revealed some discernible adjust- ments in response to KRM1 binding, especially at the bottom region near the lipid exit. The elevation of the VP2 EF-loop deforms its contacts with the VP1 gating loop, which may increase the flexibility of the gating loop around the lipid exit and thereby facilitate release of the pocket factor (SI Appendix, Fig. S6C). This local movement becomes even more significant in the structure of the KRM1-bound virion at pH 5.5 and results in a larger tunnel for pocket factor ejection (SI Appendix, Fig. S6D). Fig. 1. Binding kinetics between KRM1 and CV-A10 particles in neutral and These observations may provide an explanation for the structural acidic conditions. (A and C) SPR binding profiles of KRM1 to mature CV-A10 basis of KRM1-induced pocket factor release of CV-A10 and its virions at pH 7.4 (A) and pH 5.5 (C). The viral particles were immobilized on the chip surface, and KRM1 was applied as the analyte via serial dilutions. (B greater efficiency in low-pH conditions. In addition, this gating and D) Binding profiles of KRM1 to empty CV-A10 capsids at pH 7.4 (B) and loop becomes highly disordered in the A-particle intermediate

pH 5.5 (D). KD values are shown as the mean ± SEM of three independent and the empty capsid, impairing an important element for experiments. KRM1 interaction (SI Appendix, Fig. S6 A and B). Therefore, the

18712 | www.pnas.org/cgi/doi/10.1073/pnas.2005341117 Cui et al. Downloaded by guest on September 26, 2021 MICROBIOLOGY

Fig. 2. Cryo-EM structures of three kinds of CV-A10 particles at pH 7.4 and pH 5.5. (A) Density map of a mature CV-A10 particle at pH 7.4 (3.1 Å) (Upper)anda close-up view of the pocket region within the canyon (Lower). The density map is colored by radius, as indicated by the legend bar. The atomic model is shown in ribbons and colored by chains (VP1, blue; VP2, green; VP3, red). The “pocket factor” is shown as sticks, and the corresponding density map (mesh) is shown at 2.5σ contour. (B–F) Density maps and zoom-in views of atomic models of CV-A10 particles at different states/conditions, shown in the same style as in A.

KRM1 receptor selectively binds to the mature viral particles but To verify our structural analysis, we performed mutagenesis not to the empty capsids or A-particle intermediates (Fig. 1). studies of KRM1 by individually replacing the key interacting residues (D90, W106, and Y165) with alanine. All mutants were Interactions between the CV-A10 Viral Particle and KRM1. KRM1 purified as well-behaved proteins by size-exclusion chromatog- stands above the canyon and binds across two adjacent asym- raphy (SI Appendix, Fig. S1 D–F); however, their binding affin- metric units (ASUs) through its KR and WSC domains (Fig. 4A). ities to CV-A10 were significantly reduced compared with the Two copies of neighboring VP1s are involved in the interaction wild-type (WT) KRM1 (SI Appendix, Fig. S1 G–J), confirming with KRM1 via different regions. In the VP1 of the first ASU their critical roles in interactions with the virion. (VP1.1), the EF-loop (residues K161 and T163) and the GH gating In addition, KRM1 has also been characterized as a signal loop (residues T209, S217, and T219) form two binding patches transduction receptor on the cell surface that recognizes DKK1 with KRM1 on both sides of the canyon (Fig. 4 B and C). Meanwhile, and cooperates with LRP6, forming a ternary complex, to at- β the C terminus of VP1.2 further stabilizes the interaction by tenuate Wnt/ -catenin signaling activation (7, 8). Compared with contacting the WSC domain of KRM1 (Fig. 4D). In addition, the CV-A10/KRM1 complex, we found that KRM1 uses a highly the KR domain of KRM1 covers the VP2.1 EF-loop protrusion similar interface to bind CV-A10 or DKK1/LRP6, suggesting that these interactions are mutually exclusive (Fig. 4 G–J and (residues 138 to 143) and contributes the majority of strong SI Appendix charged interactions (Fig. 4E). Specifically, residues D88, D90, , Fig. S8). and Y105 in the KR domain hydrogen-bond to the amidogen of KRM1-Mediated CV-A10 Uncoating at Acidic pH In Vitro. Because the K140 and N142 and to the carbonyl of T139, respectively. The acidic pH conditions enable efficient pocket factor release of side chains of W94 and W106 in the KR domain contact K140 CV-A10 on KRM1 interaction, we sought to investigate whether π of VP2.1 through -cation interactions. Moreover, N140 and soluble KRM1 could directly induce CV-A10 virion uncoating Y165 form additional hydrogen bond and hydrophobic inter- in vitro. To do so, we incubated mature CV-A10 particles with actions with VP3.2 (Fig. 4F and SI Appendix,TableS2). Of soluble KRM1 in both neutral (pH 7.4) and acidic (pH 5.5) note, K140 in VP2.1 and residues 180 to 182 (TGG) in VP3.2 conditions at physiological temperature (37 °C) and compared (the key residues participating in receptor binding) are con- the virion morphologies by negative stain electron microscopy (EM). served among all KRM1-dependent EV-As identified so far (SI The mature CV-A10 particles alone did not exhibit significant changes Appendix,Fig.S7), indicating a potentially universal mechanism under these treatments, consistent with the cryo-EM structures for receptor binding by these related viruses. (Fig. 2). In the presence of KRM1, a substantial proportion of

Cui et al. PNAS | August 4, 2020 | vol. 117 | no. 31 | 18713 Downloaded by guest on September 26, 2021 Fig. 3. KRM1-bound CV-A10 structures under neutral and acidic conditions. (A) Density maps of the mature CV-A10/KRM1 complex at pH 7.4 (3.0 Å; Left)and pH 5.5 (3.0 Å; Right) viewed along the icosahedral two-fold axes. (Insets) Close-up views of asymmetric units at the receptor-binding interface. (B) Close-up view of the hydrophobic pocket and the “pocket factor” inside, corresponding to pH 7.4 and pH 5.5, respectively. The density map is shown at 2.5σ contour. (C) Atomic models of an asymmetric unit of a CV-A10 virion with (dark colors) or without (light colors) KRM1 binding are aligned. The VP2 EF-loop, VP3 C terminus, and GH-loop are highlighted by dashed boxes, and their close-up views are shown at the side. KRM1 is not shown in the superimposition for clarity. (D) Comparison of each viral subunit in the CV-A10/KRM1 complex at pH 7.4 and pH 5.5. The structures at neutral (gray) and acidic (colored) pH conditions are shown as ribbons and overlaid for each subunit. The N and C termini of each protein are indicated by black dots.

viral particles transformed into empty capsids under acidic trigger uncoating in acidic conditions (3, 12). A special example pH conditions, whereas no discernible changes occurred at among the EV-Cs is poliovirus, which uses CD155 (PVR) as a pH 7.4 (Fig. 5). bifunctional receptor for both attachment and uncoating (5). As These observations demonstrate that KRM1 is a two-in-one for EV-As, the scavenger receptor class B member 2 (SCARB2) receptor for CV-A10 entry that can mediate both viral attach- has been identified as the receptor for EV-A71 and CV-A16 (13). ment and uncoating to enable virus entry. Interestingly, the low- A recent structural study also revealed an unexpected binding pH condition is much more favorable than neutral conditions for mode between EV-A71 and SCARB2, where SCARB2 binds to viral uncoating, suggesting that the viruses may infect cells the “southern rim” of the canyon, interacting with VP1 and VP2 mainly through the endosomal pathway. As reported previously, to initiate uncoating (SI Appendix,Figs.S8B and S10). Sequence the binding of KRM1 to CV-A10 would trigger the endocytosis alignment for SCARB2-dependent EV-As shows low conserva- of viral particles into the endosome (6), which would then fa- tion in the SCARB2-binding residues (14), suggesting remarkable cilitate the uncoating process in the acidic environment. In ad- variability for interactions with the same receptor. In this study, we dition, as the binding of KRM1 to CV-A10 at neutral conditions found that KRM1 binds across two ASUs and interacts with all resulted in low-efficiency release of the pocket factor, the un- viral proteins in the outer shell, for which the binding interface is coating process may also occur at the cell membrane as a minor moderately conserved among all KRM1-dependent EV-As char- alternative pathway for cell entry (Fig. 6). acterized so far (SI Appendix,Fig.S7). These findings suggest that KRM1 displays a more uniform recognition pattern by different Discussion EV-As. Enteroviruses infect host cells by recognizing membrane receptors, As observed in many enteroviruses, uncoating receptors usu- followed by uncoating at either the cell surface or in the endosome ally bind to the canyon of the virion via their Ig-like domain and (11). Some of these viruses bind to different receptors for attach- trigger pocket factor release to initiate conformational changes ment and uncoating, whereas several viruses use a single receptor for uncoating (3, 11). However, KRM1 binds to CV-A10 above for both steps. Many EV-Bs use CD55 for attachment to the the canyon through its KR and WSC domains instead of through cell surface and switch to the endosomal receptor FcRn to the CUB pseudo Ig-like domain (SI Appendix,Fig.S9A). It covers

18714 | www.pnas.org/cgi/doi/10.1073/pnas.2005341117 Cui et al. Downloaded by guest on September 26, 2021 MICROBIOLOGY

Fig. 4. Molecular interaction between CV-A10 and KRM1. (A) The overall contacting interface between KRM1 and the CV-A10 virion. Two ASUs of viral proteins are shown in surface models (asu.2 in light colors). VP1, VP2, and VP3 of CV-A10 are shown in blue, green, and red, respectively. KRM1 includes three domains—KR (black), WSC (orange), and CUB (violet)—and is represented as ribbons. (B–F) Interactions between the CV-A10 virus and KRM1 receptor. The proteins are colored by chains with the same color code as in A. The key interacting residues are displayed as sticks and colored by elements. Hydrogen bonds are represented by dashed lines. (G and I) Comparison of the binding mode between CV-A10/KRM1 and the DKK1-LRP6/KRM1 complex. CV-A10, DKK1, and LRP6 are drawn in surface representation. (H and J) KRM1 from the CV-A10/KRM1 complex and the DKK1-LRP6/KRM1 complex is shown in surface representation and colored in gray. The CV-A10, DKK1, and LRP6 binding interfaces are highlighted with the same colors as the interacting protein (VP1.1, blue; VP2.1, green; VP3.1, red; VP1.2, light blue; VP3.2, salmon; DKK1, pale yellow; and LRP6, slate).

the canyon on both sides but does not penetrate inside to directly increase its flexibility. The role of histidine residues serving as interact with the canyon, which is unique compared with the more pH sensors is well established for the fusogenic glycoproteins of prevalent canyon-penetrating receptors. It is interesting that KRM1 enveloped viruses (15), and it may also apply to the pH-dependent interacts with the distal region of the gating loop to potentially alter entry processes of nonenveloped viruses. In addition, temperature its conformation around the lipid exit within the canyon, thereby may affect the stability of virion–receptor interactions, further facilitating release of the “pocket factor.” How the acidic environ- modulating the uncoating efficiency. In the cryo-EM images, the ment triggers more significant conformational changes of the gating relative abundance of empty capsids and A-particles did not loop to accelerate pocket factor release remains unclear. A possible exhibit obvious changes in the presence of KRM1, irrespective of mechanism is that the low pH condition leads to protonation of the pH condition (SI Appendix,Fig.S2). This observation suggests histidine residues around the canyon, which may perturb the in- that KRM1 binding may temporarily stabilize the virion structure teractionnetworksofthegatinglooptoitsanchorpartnersand after pocket factor release at low temperature (4 °C in this case),

Cui et al. PNAS | August 4, 2020 | vol. 117 | no. 31 | 18715 Downloaded by guest on September 26, 2021 the role of KRM1 for mediating entry by CV-A10 and related enteroviruses (Fig. 6). It has been shown that viruses tend to use the cell signaling and regulatory pathways to induce endocytosis and facilitate subsequent invasion (17, 18). On the other hand, KRM1 is also involved in the conserved Wnt/β-catenin pathway, in which it interacts with DKK1 and LRP6 to form a ternary complex and induce endocytosis. Previous studies have shown that KRM1 binds to DKK1-LRP6 via its KR and WSC domains, and the introduction of N-linked glycans on D90 also decreases the binding capacity of KRM1 to DKK1 (7). The key interacting residues involved in the CV-A10–KRM1 interaction significantly overlap with the interface for DKK1/LRP6 binding. Thus, CV-A10 could possibly mimic DKK1-LRP6 for KRM1 binding and internalization to enable cell entry. In addition, a previously reported CV-A10/Fab complex structure revealed that the VP1 C terminus, VP2 EF-loop, and VP3 AB- loop constitute candidate neutralization epitopes (1, 19) (SI Appendix,Fig.S9C). Here we found that the VP2 EF-loop (residues 138 to 143) contributed the majority of the strong charged interactions with KRM1, suggesting that this region might be a hotspot of interference with the receptor-binding process to block virus entry. In addition, the KRM1-binding interface on CV-A10 covers a broad region with multiple contacting patches, which may imply potent epitopes for developing vaccines and therapeutic antibodies. In summary, here we present the molecular basis for KRM1- mediated CV-A10 entry into host cells and propose that KRM1 is a two-in-one attachment and uncoating receptor. Attachment at neutral pH enables the virus association with cells at the plasma membrane, and the low-pH conditions in the endosome accelerate pocket factor release to initiate the uncoating process. These findings provide systematic mechanistic insight into the entry of CV-A10 and related viruses, which may constitute an important basis for developing antiviral therapeutics.

Fig. 5. KRM1-dependent uncoating of CV-A10 in vitro. (A–D) Representa- tive negative stain EM images of four CV-A10 virus groups without (A and B) or with (C and D) KRM1 proteins and exposed to neutral pH (pH 7.4, Upper Left) or acidic pH (pH 5.5) treatments. Samples were negatively stained immediately after mixing or after a 10-min incubation at 37 °C. (E) The per- centage of empty particles for each group is displayed in the interleaved scatterplot, with a horizontal line indicating the median. The significance of differences was tested by Tukey’s multiple comparison test. ****P < 0.0001; ns, not significant.

thus preventing conformational changes of the viral capsid to initiate genome release. In contrast, the receptor would be prone to dissociate at physiological temperature from the metastable intermediate, resulting in efficient genome release from the virions (Fig. 5). During the preparation of this manuscript, Zhao et al. (16) reported a similar near-atomic resolution (3.9 Å) structure of KRM1 bound to CV-A10 at pH 8 and suggested that KRM1 could potentially induce pocket factor release of the virion. This is consistent with our observation that KRM1 binding slightly reduced the occupancy of pocket factor in the density map at pH 7.4 compared with the free virion. However, the authors did not Fig. 6. KRM1 is a two-in-one receptor for CV-A10 entry into the cell. KRM1 investigate the effects of different pH conditions on virion–receptor is anchored on the cell membrane, and the three structural domains (KR, interactions and thus could not comprehensively define the mech- WSC, and CUB) are displayed in red, yellow, and slate, respectively. The four structural proteins (VP1, VP2, VP3, and VP4) within the viral capsid are in- anism of virus entry. Moreover, our structures and biochemical dicated in violet purple, brown, chocolate, and sky blue, respectively. The evidence allowed better resolution of the mechanistic details at viral RNA genome and the “pocket factor” are shown in forest green and different stages of virus entry and provided more insight into orange, respectively.

18716 | www.pnas.org/cgi/doi/10.1073/pnas.2005341117 Cui et al. Downloaded by guest on September 26, 2021 Materials and Methods with Gautomatch (by K. Zhang). The particles were extracted as 4× binned Virus Production and Purification. The CV-A10 strain HB09-035 was first iso- particles and subjected to reference-free 2D classification with Relion 3.0 (22). lated in 2009 in Hebei Province, China. Human RD cells were maintained in Ice contamination, damaged particles, and low-contrast images were re- Dulbecco’s modified Eagle’s medium (Invitrogen) supplemented with 10% moved after multiple rounds of classification and subset selection. The clean dataset was recentered and reextracted for subsequent 3D classification with fetal bovine serum (Invitrogen) and 100 U/mL penicillin-streptomycin. RD the density map of free CV-A10 mature virions (EMD-9674) as the reference cells were grown to 95% confluence and infected with CV-A10 at a multi- model, which was low-pass filtered to 60 Å before image alignment. The plicity of infection (MOI) of 0.1. Infected cells were incubated at 37 °C for 36 h, angular assignment was initiated with a global search and proceeded with a at which point ∼90% of cells displayed a cytopathic effect. The cultures were restricted local search based on the estimation of alignment accuracy. For harvested and centrifuged at 12,000 × g for 60 min to remove cell debris. each dataset, viral particles in different states were observed with clear fea- Viruses from the supernatant were purified as described previously (3). In tures of secondary structural elements. The most distinguished classes with brief, the virus was precipitated with a 30% (wt/vol) sucrose cushion and the best alignment accuracy were selected for 3D refinement for each struc- resuspended in phosphate-buffered saline (PBS). Two bands were observed ture. After the initial refinement with 2× binned particles, the full-size particle on the 15 to 45% (wt/vol) sucrose density gradient. The sample was frac- images were further recentered and reextracted for a second round of 3D tionized into aliquots of 200 μL from the top to the bottom of the sucrose refinement. At the final stage, CTF refinement was performed with Relion cushion. Each fraction was quantified by measuring the optical density 3.0 (22) to counteract the defocus variation of each particle, and the dose- (OD) at 260 and 280 nm with a NanoDrop spectrophotometer and further weighted images were calculated with MotionCor2 (20) to balance the visualized by negative stain EM and sodium dodecyl sulfate–polyacrylamide gel effects of radiation damage. An additional round of 3D refinement was electrophoresis. The top band was composed mainly of empty capsid particles performed with per-particle CTF values and dose-weighted images to cal- (OD /OD = 0.9), and the bottom band consisted mainly of mature virions 260 280 culate the final reconstruction. The numbers of initial and final particles (OD /OD = 1.7) (SI Appendix,Fig.S1A and B). Then the two bands were 260 280 and the composition of different conformational particles in each dataset individually dialyzed against 1×PBS (pH 7.4) buffer. are shown in SI Appendix, Table S1 and Fig. S2. Local resolution was estimated using ResMap (23) for each reconstruction. Protein Production and Purification. The construct of soluble human KRM1 protein has been described previously (7). In brief, the human KRM1 ecto- Model Building and Refinement. The crystal structure of KRM1 (Protein Data domain (A23-G373) was synthesized and cloned into pCMV3 (Sino Biological) Bank [PDB] ID code 5FWS) and the cryo-EM structure of CV-A10 (PDB ID code with a human IL-2 signal sequence at the N terminus and 10 histidines at the 6ACU) were used as the initial model to fit the CV-A10/KRM1 density map C terminus using the HindIII and BamHI sites, respectively. The recombinant (pH 7.4, 3.0 Å; pH 5.5, 3.0 Å) using Chimera (24). After initial fitting, the plasmid was transiently transfected into HEK293T cells for KRM1 expression. model was manually adjusted in Coot (25) to improve map fitting and up- The supernatant was harvested at 7 d posttransfection and centrifuged to date the sequence registers. The density map of an ASU was then segmented remove cell debris, and the protein was subsequently purified by Ni-nitrilotriacetic using Chimera (24) and placed into a pseudo-crystallographic unit cell (P1

acid chromatography (GE Healthcare Life Sciences). The target proteins were space group) for model refinement. The atomic model was refined against MICROBIOLOGY eluted using 300 mM imidazole dissolved in PBS. The fractions were then dialyzed the ASU map in real space using Phenix (26), with secondary structure re- into PBS, followed by size-exclusion chromatography using a Superdex 200 In- straints applied. Model statistics (including bond lengths, bond angles, all- crease 10/300 GL column (GE Healthcare Life Sciences). The human KRM1 atom clash, and rotamer outliers), and Ramachandran plot statistics were mutant plasmids were constructed by site-directed mutagenesis, for which monitored during the refinement procedure. After several rounds of itera- the identified key residues involved in CV-A10 binding (D90A, W106A, and tive model building and refinement, the model coordinates converged and Y165A) were mutated separately, and the proteins were purified by the fit well in the density map by visual inspection. At this point, the individual B – same methods (SI Appendix,Fig.S1C F). factors and occupancies of each atom were refined using the standard re- ciprocal space refinement procedure in Phenix (26). The stereochemical SPR Assay. The binding kinetics and affinity of KRM1 to CV-A10 were ana- quality of the final model was assessed using MolProbity (27). The statistics lyzed by SPR using a BIAcore 3000 biosensor (GE Healthcare Life Sciences) at for 3D reconstruction and model refinement are summarized in SI Appendix, room temperature (25 °C). All proteins used for kinetic analyses were ex- Table S1. changed into PBST buffer through gel filtration. Mature CV-A10 virus was biotinylated and immobilized on a SA chip to approximately 6,000 response Structure Analysis and Visualization. The reconstructed maps and atomic units. The soluble ectodomain of KRM1 and the D90A, W106A, Y165A mu- models were visualized using Chimera (24) and analyzed using the wrapped tant proteins were serially diluted with PBST, sequentially injected into the applications. All EM density figures were rendered with Chimera, and atomic μ chip at a flow rate of 30 L/min for 1 min, and then allowed to dissociate model representations were generated with the PyMOL molecular graphics over 2 min. The binding curve at zero concentration of protein was sub- system (https://pymol.org/2/). tracted as a blank from each experimental curve. After each cycle, a short injection of 3 M MgCl2 was used to regenerate the sensor surface. Data were In Vitro Uncoating Assays and Negative Stain EM. Purified mature CV-A10 virus analyzed, and kinetic constants were estimated using BIAevaluation 4.1 was incubated alone or mixed with KRM1 protein at a molar ratio of 1:70 in software (GE Healthcare Life Sciences). pH 7.4 or pH 5.5 buffers. Samples were negatively stained immediately after mixing or after a 10-min incubation at 37 °C. Grids were examined with an Cryo-EM Sample Preparation and Data Collection. Purified mature CV-A10 FEI Tecnai G20 transmission electron microscope operated at an accelerating virus was incubated with excess KRM1 at final concentrations of 2 mg/mL voltage of 120 kV. Images were recorded at a magnification of 29,000×, and 0.5 mg/mL for 10 min at 4 °C. The CV-A10 viral particle or CV-A10/KRM1 resulting in a pixel size of 3.76 Å/pixel. Defocus values varied from −2to−4 μm. complex samples (3 μL), at pH 7.4, were placed on a freshly plasma-cleaned Each image was exposed with a total dose of 60 e−/Å2 and recorded with a BM- lacey carbon grid and allowed to adsorb to the grid for 60 s, followed by Eagle CCD camera. The empty particles were quantified by visual inspection. blotting with a filter paper for 3 s before plunge-freezing using a Vitrobot The statistics for each sample were calculated with GraphPad Prism using Mark IV (FEI) operated at 4 °C and 100% humidity. For the acidic condition approximately 2,000 total particles from randomly selected images. The samples, CV-A10/KRM1 samples were incubated in low pH buffer [20 mM significance of difference was tested by Tukey’s multiple comparison 2-(N-morpholino) ethanesulphonic acid and 150 mM NaCl, pH 5.5] for implemented in GraphPad Prism. 10 min, then applied onto the grid as described above. Cryo-EM data were collected using either a 300-kV FEI Titan Krios (for CV- Data Availability. The density maps have been deposited to the Electron A10 alone) or a 200-kV FEI Talos Arctica (for CV-A10/KRM1 complex) electron Microscopy Data Bank with accession codes EMD-30253 (mature CV-A10, pH 7.4), microscope equipped with a Gatan K2 Summit direct electron detector. EMD-30254 (mature CV-A10, pH 5.5), EMD-30287 (A-particle CV-A10, pH 7.4), Images were recorded in superresolution counting mode with a calibrated EMD-30290 (A-particle CV-A10, pH 5.5), EMD-30291 (empty CV-A10, pixel size of 1.35 Å /1.32 Å. Defocus values varied from -1.5 to −2.5 μm. Each pH 7.4), EMD-30292 (empty CV-A10, pH 5.5), EMD-30259 (CV-A10/KRM1, − 2 image was dose-fractionated to 32 frames with a total electron dose of 40 e /Å . pH 7.4), and EMD-30260 (CV-A10/KRM1, pH 5.5). The coordinates of the corresponding atomic models were deposited to the Protein Data Bank Image Processing. The beam-induced image drift and anisotropic magnifi- with ID codes 7BZN (mature CV-A10, pH 7.4), 7BZO (mature CV-A10, cation of each image stack were corrected with the MotionCor2 program (20). pH 5.5), 7C4T (A-particle CV-A10, pH 7.4), 7C4W (A-particle CV-A10, pH 5.5), Initial parameters of the contrast transfer function (CTF) were determined by 7C4Y (empty CV-A10, pH 7.4), 7C4Z (empty CV-A10, pH 5.5), 7BZT (CV-A10/ CTFFIND4 (21) at the micrograph level. Particles were automatically picked KRM1,pH7.4),and7BZU(CV-A10/KRM1,pH5.5).

Cui et al. PNAS | August 4, 2020 | vol. 117 | no. 31 | 18717 Downloaded by guest on September 26, 2021 ACKNOWLEDGMENTS. We thank Dr. Yong Zhang (Chinese Center for Disease experiments. This work was supported by the National Science and Technology Control and Prevention) who kindly provided the CV-A10 HB09-035 strain Major Projects (2018ZX09711003), the Strategic Priority Research Program of (GenBank accession no. MT263729). We thank all staff members in the Center the CAS (XDB29010202), the National Science and Technology Major Project of Biological Imaging, Institute of Biophysics, Chinese Academy of Sciences (2018ZX10733403), the Technological Innovation Project of Shanxi Transforma- (CAS), Beijing, and the Cryo-EM Center of Southern University of Science and tion and Comprehensive Reform Demonstration Zone (2017KJCX01), the Beijing Technology for assistance with cryo-EM data collection. We are grateful to the staff members in the EM department of the State Key Laboratory of Nova Program of Science and Technology (Z191100001119030), and the Youth Membrane Biology, Institute of Zoology, CAS, Beijing, for technical support in Innovation Promotion Association of the CAS (20200920). R.P. is supported by the electron microscope operation. Special thanks to Drs. Zheng Fan and Wei Young Elite Scientist Sponsorship Program of the China Association for Science Zhang of the Institute of Microbiology, CAS for assisting with the SPR and Technology (2018QNRC001).

1. R. Zhu et al., Discovery and structural characterization of a therapeutic antibody 15. R. Peng et al., Structures of human-infecting Thogotovirus fusogens support a com- against coxsackievirus A10. Sci. Adv. 4, eaat7459 (2018). mon ancestor with insect baculovirus. Proc. Natl. Acad. Sci. U.S.A. 114, E8905–E8912 2. L. Bian et al., Hand, foot, and mouth disease associated with coxsackievirus A10: More (2017). serious than it seems. Expert Rev. Anti Infect. Ther. 17, 233–242 (2019). 16. Y. Zhao et al., Hand-foot-and-mouth disease virus receptor KREMEN1 binds the 3. X. Zhao et al., Human neonatal Fc receptor is the cellular uncoating receptor for canyon of Coxsackie Virus A10. Nat. Commun. 11, 38 (2020). Enterovirus B. Cell 177, 1553–1565.e16 (2019). 17. P. Cossart, A. Helenius, Endocytosis of viruses and bacteria. Cold Spring Harb. Per- 4. M. G. Rossmann, Y. He, R. J. Kuhn, Picornavirus-receptor interactions. Trends Micro- spect. Biol. 6, a016972 (2014). biol. 10, 324–331 (2002). 18. K. M. Hussain, K. L. Leong, M. M. Ng, J. J. Chu, The essential role of clathrin-mediated 5. M. Strauss et al., Nectin-like interactions between poliovirus and its receptor trigger endocytosis in the infectious entry of human enterovirus 71. J. Biol. Chem. 286, – conformational changes associated with cell entry. J. Virol. 89, 4143–4157 (2015). 309 321 (2011). 6. J. Staring et al., KREMEN1 is a host entry receptor for a major group of enteroviruses. 19. W. Dai et al., Recombinant virus-like particle presenting a newly identified coxsackievirus A10 neutralization epitope induces protective immunity in mice. Antiviral Res. Cell Host Microbe 23, 636–643.e5 (2018). 164, 139–146 (2019). 7. M. Zebisch, V. A. Jackson, Y. Zhao, E. Y. Jones, Structure of the dual-mode wnt reg- 20. S. Q. Zheng et al., MotionCor2: Anisotropic correction of beam-induced motion for ulator kremen1 and insight into ternary complex formation with LRP6 and Dickkopf. improved cryo-electron microscopy. Nat. Methods 14, 331–332 (2017). Structure 24, 1599–1605 (2016). 21. A. Rohou, N. Grigorieff, CTFFIND4: Fast and accurate defocus estimation from elec- 8. B. Mao et al., Kremen proteins are Dickkopf receptors that regulate Wnt/beta-catenin tron micrographs. J. Struct. Biol. 192, 216–221 (2015). signalling. Nature 417, 664–667 (2002). 22. D. Kimanius, B. O. Forsberg, S. H. W. Scheres, E. Lindahl, Accelerated cryo-EM struc- 9. J. Chen et al., Coxsackievirus A10 atomic structure facilitating the discovery of a ture determination with parallelisation using GPUs in RELION-2. Elife 5, e18722 broad-spectrum inhibitor against human enteroviruses. Cell Discov. 5, 4 (2019). (2016). 10. L. Zhu et al., Structures of Coxsackievirus A10 unveil the molecular mechanisms of 23. A. Kucukelbir, F. J. Sigworth, H. D. Tagare, Quantifying the local resolution of cryo-EM receptor binding and viral uncoating. Nat. Commun. 9, 4985 (2018). density maps. Nat. Methods 11,63–65 (2014). 11. A. I. Wells, C. B. Coyne, Enteroviruses: A gut-wrenching game of entry, detection, and 24. E. F. Pettersen et al., UCSF Chimera–A visualization system for exploratory research evasion. Viruses 11, 460–479 (2019). and analysis. J. Comput. Chem. 25, 1605–1612 (2004). 12. S. Niu et al., Molecular and structural basis of Echovirus 11 infection by using the dual- 25. P. Emsley, B. Lohkamp, W. G. Scott, K. Cowtan, Features and development of Coot. receptor system of CD55 and FcRn. Chin. Sci. Bull. 65,67–79 (2020). Acta Crystallogr. D Biol. Crystallogr. 66, 486–501 (2010). 13. S. Yamayoshi et al., Scavenger receptor B2 is a cellular receptor for enterovirus 71. 26. P. D. Adams et al., PHENIX: A comprehensive Python-based system for macromolec- Nat. Med. 15, 798–801 (2009). ular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66, 213–221 (2010). 14. D. Zhou et al., Unexpected mode of engagement between enterovirus 71 and its 27. V. B. Chen et al., MolProbity: All-atom structure validation for macromolecular crys- receptor SCARB2. Nat. Microbiol. 4, 414–419 (2019). tallography. Acta Crystallogr. D Biol. Crystallogr. 66,12–21 (2010).

18718 | www.pnas.org/cgi/doi/10.1073/pnas.2005341117 Cui et al. Downloaded by guest on September 26, 2021